Challenging Recognition Evidence from Videos: R v Fitzpatrick, 2026 ONCA 262

Resemblance evidence from a “stumble” excluded and a new trial ordered.

When police or Crown witnesses claim they can identify your client from surveillance footage, defence counsel should be reaching for Fitzpatrick. The ONCA’s April 2026 decision draws a sharp line between admissible recognition evidence and inadmissible resemblance evidence and clarifies when the distinction has real teeth at trial.

The Facts in Brief

Two men were captured on surveillance video approaching and leaving the scene of a fatal shooting. The Crown alleged one was Fitzpatrick, an accomplice. In support, it called Fitzpatrick’s father and stepmother, neither of whom could actually identify Fitzpatrick in the footage. What they could say was that Person #2’s stumble over a curb reminded them of Fitzpatrick because of his clumsiness. The trial judge admitted this evidence under a Leaney framework. Fitzpatrick was convicted of manslaughter.

“A non-distinctive stumble or trip over a curb cannot reasonably amount to evidence of a suspect resembling the accused.”

The Legal Framework: Recognition vs. Resemblance

Recognition evidence is a form of lay opinion evidence. The test, drawn from Berhe (2012 ONCA 716), is whether: (1) the witness is sufficiently familiar with the accused to have some basis for an identification opinion; and (2) the witness is in a better position than the trier of fact to identify the person in the footage.

The ONCA held that this threshold simply was not met. Neither witness could identify anyone in the video. If they couldn’t identify Fitzpatrick, they couldn’t recognize him. Their familiarity with him was “ultimately meaningless” because it didn’t help them do the one thing the evidence was supposed to accomplish.

Critically, the Court distinguished resemblance evidence, which it described as evidence of something specific about a suspect resembling the accused, such as a distinctive physical feature, a characteristic gait, or particular clothing, from the stumble evidence here, which was a generic, non-identifying observation. Probative value in this context comes from specificity. A random trip over a curb has none.

Why This Matters for the Police Officer Scenario

The police officer identification scenario is common: an investigator who has interacted with an accused, reviewed their photos, or simply studied the video at length concludes that the person in the footage is your client. Fitzpatrick gives you several lines of attack:

• Push for definiteness. The Court emphasized that neither witness “purported to definitively identify anyone in the video.” If an officer’s identification is hedged, qualified, or based on impression rather than recognition, that goes to the threshold admissibility question, not just weight.
• Disaggregate the evidence. The trial judge admitted the evidence as a “constellation of factors” — clothing, build, and the stumble together. The ONCA pulled that apart: the stumble evidence had no probative value on its own, and clothing (a Raptors hoodie, generic shoes) had “limited probative value given the ubiquity of those items.” Make the Crown justify each component.
• Object to the stumble and similar non-gait behaviour. The trial judge specifically found a stumble is not a characteristic of someone’s gait. Non-distinctive movements, nervous behaviours, or generic mannerisms captured on video cannot serve as the basis for resemblance or recognition evidence.
• Police familiarity is not the same as genuine prior acquaintance. The Berhe “prior acquaintance/better position” test contemplates witnesses who know the accused in a real sense. A police officer who has reviewed booking photos or watched footage repeatedly may not be in a meaningfully better position than the jury — and arguably is worse, given the risk of confirmation bias.
• Watch for Crown bootstrapping. In Fitzpatrick, the Crown used the witnesses’ reluctance to testify consistently with their police statements as evidence of the importance of their initial reactions. The ONCA noted how the stumble evidence became “a central feature of the Crown’s case” despite having “no bearing on the issue of Person #2’s identity.” Object early and renew your objection at closing if Crown counsel overreaches.

Jury Instructions: A Second Front

The Court also ordered a new trial for co-accused Wright on two grounds: the trial judge failed to include the eyewitness C.H.’s exculpatory description of the shooter as a Black man in the W.(D.) instructions, and Crown counsel’s closing address improperly relied on Miaponoose to suggest eyewitness evidence is inherently unreliable, then personalized it by telling the jury this is something all criminal lawyers know. The corrective instruction, delivered five days later, was found insufficient.

The takeaway for jury trials: where a Crown witness gives evidence that could raise a reasonable doubt, even if they don’t fully support the defence, ensure the W.(D.) instruction captures it. And if Crown counsel steps outside the record in closing, press hard for a mistrial. The ONCA found the curative instruction fell short here precisely because Crown counsel had framed the improper remarks as insider knowledge the jury was being let in on.

Bottom Line

Fitzpatrick is a useful authority any time the Crown seeks to use lay opinion evidence to place your client at a scene via surveillance footage. The prior acquaintance/better position test has real content. Witnesses who hedge, who cannot make a definitive identification, and who point to non-distinctive behaviour as their basis for “recognition” are not providing admissible recognition evidence, they’re providing inadmissible resemblance evidence dressed up as something more. The distinction matters.

R v Fitzpatrick, 2026 ONCA 262 (Rahman JA, Roberts and Pomerance JJA concurring). Appeal heard June 11, 2025; judgment April 14, 2026. New trials ordered for both appellants.